Proposed-site record
High-priority Google/Panattoni Project Cannoli proposal in Van Buren Township. Township planning records document the preliminary site-plan path for three data-processing buildings and a substation; the campus-scale power lead remains separate from DTE's broader generation partnership and from confirmed IT load.
This panel tracks what the facility is for without assuming that every project is AI training, inference, or cloud capacity.
No source-backed workload, customer, hardware, or training/inference allocation is attached to this record yet.
Workload classification remains unknown in the reviewed record.Source leads point to attached records or reporting. They still need field-level review before being treated as verified evidence.
No public funding record attached yet. Local subsidy, tax-abatement, land, bond, and utility records need follow-up.
Source reviewed September 9, 2026: Van Buren Township's preliminary site-plan materials identify three data-processing buildings, a network-service building, a water-services building, and stormwater ponds, but the reviewed public record does not establish a water-service quantity, cooling withdrawal, consumption, or wastewater capacity.
Source reviewed September 9, 2026: Van Buren Township's February 11, 2026 agenda records the preliminary site-plan request for three data-processing buildings and a substation; Panattoni describes a Fortune 50 cloud/data-processing use, while reporting identifies Google as the end user. DTE's separate 2.7 GW clean-resource partnership is grid-generation context, not this campus's load. The dataset's 1,000-2,700 MW remains proposed project/power context, not confirmed requested, approved, energized, generation, or data-center IT-load capacity.
No fuel-cell or Bloom Energy source is attached to this project yet.
No NDA reported
Start a moderated thread for local questions, meeting notes, public-records coordination, or context that should not be treated as verified evidence yet.
These links are the current public trail for this record. Treat broad dataset links as leads; local documents and reporting carry more weight.